Genomic patterns in low-grade prostate cancer indicative of upgrading


Session type:

Salpie Nowinski1,Jelmar Quist1,Trevor Graham2,Massimo Loda3,Mieke Van Hemelrijck1,Anita Grigoriadis1
1King's College London,2Queen Mary University London,3Dana-Farber Cancer Institute



Active surveillance (AS) for men with low-risk prostate cancer (PCa), aims to reduce over-treatment. Due to a lack of predictive disease progression markers, most current protocols for AS depend on repeated biopsies to determine eligibility to stay on AS. We aimed to identify molecular patterns indicative of upgrading in PCa for men on AS.


TAPS2.0 absolute copy number was established for 45 Gleason 6 (3+3), 145 Gleason 7 (3+4), and 100 Gleason 7 (4+3) for TCGA Affymetrix SNP6.0. Fisher’s exact tests determined regions of significant copy number differences.


Gleason 6 PCas display significantly less frequent gains (t-test p=0.001), losses (p=0.0002), and cnLOHs (p=0.0003) than Gleason 7 PCas. To identify distinctive groups among Gleason 6 PCa genomes, a two-way hierarchical clustering using categorised copy number data for 811 cytobands for all low Gleason PCa was performed. A cluster enriched for genomically stable Gleason 6 samples was identified encompassing 16/44 (36%) samples, while the remaining Gleason 6 PCas were grouped into two clusters and resembled the majority of Gleason 7 (4+3) and (3+4) samples, exhibiting unstable cancer genomes. By dichotomising all Gleason 6 samples based on this clustering, significant regions with different copy number aberrations were identified. These included recurrent losses with a higher frequency in the genomically unstable Gleason 6 samples, on 8p23 (p=2.19E-05, SOX7), 13q13-q14 (p=0.016, ELF1), and 6q21 (p=0.017, FOXO3). Matched RNAseq data identified 40 genes, which were both differentially expressed and in regions of significant loss, between these two Gleason 6 samples (e.g., SOX7, FDR adj p-value <0.2).


At least two groups among Gleason 6 PCas with genomic aberrations and concordant gene expression pattern can be identified. Current cross-validation of these molecular features in two independent low-grade PCa cohorts with information of upgrading of PCa will demonstrate their potential as predictive markers for upgrading in low Gleason PCa.